To optimize strategies and measurement parameters for quantification of small fat and water fractions (<10%) in mixtures of both components by 4-point in-phase and opposed-phase gradient-echo imaging and to compare theoretical results with in-vitro experiments using emulsions. Theoretical analysis was based on steady-state signal equations for spoiled GRE-sequences and on relaxation properties of water and fat components. For quantification, signals were corrected for T2*-decay, T1-decay, and signal contributions from double bonds. Theoretical results were exemplarily compared to measurements at 1.5T on emulsions with either low water or fat fractions (0.5-10%) using spoiled 2D- and 3D-GRE-sequences. Excitation flip angle was varied in order to determine suitable values for sensitive detection of small fat/water fractions. Theoretical results and measurements correlated well, especially for 3D-sequences. Maximal sensitivity to a small signal fraction (S (fat) and S (water), respectively), was provided at the Ernst angle of the lower concentrated component. For 2D-sequences, the nominal flip angle had to be increased for compensation of slice profile effects and B(1) inhomogeneities. IP- and OP-echoes are recommended to be acquired in separate measurements with smallest possible receiver bandwidth to increase SNR/unit-time. Lowest detectable fat/water concentration in emulsions under typical conditions regarding spatial resolution and measuring time was approximately 1%. Using IP/OP-imaging with optimized parameters and post-processing, a sensitive and reliable detection of small fat/water fractions larger than 1% is possible in emulsions.